With the rapid development of the computer technology and the network technology, the Internet is playing a more and more important role in daily life, study, and work of people. A variety of applications on the Internet emerge one after another.
User generated content (UGC) is a new manner for a user to use the Internet, with which a conventional case where downloading predominates is changed to a case where downloading and uploading are of equal importance. Social networks, video sharing, microblog, and blog all are main application forms of the UGC. With the continuous development of global Internet services, a UGC service is increasingly rising, and attracts extensive attention from the industry.
Microblog is a typical UGC application, and is a user relationship based platform for sharing, disseminating, and acquiring information. A user can publish information about personal thoughts and doings in brief words (generally less than 140 words) on microblog by using various terminals, implementing instant sharing of the information. Microblog is a new manner in which a user shares and disseminates information on the Internet. Due to the characteristic of a weak relationship chain of microblog, a message on microblog is transferred very quickly, which induces many users to publish on microblog some messages adverse to the ecological development of microblog, for example, attaching their own advertisements when reposting and commenting on a message of an eminent person. Such reposts and comments that are irrelevant to a mentioned user all enter an inbox of the mentioned user, causing enormous harassment to the user, lowering the message quality of microblog and the creative enthusiasm of users, and bringing a tremendous side effect to the positive development of the microblog.
In the existing technology, a policy of processing after occurrence is adopted for a harassing mention behavior in a UGC service, that is, only when a user reports a malicious mention by an account does an examiner determine whether the account causes harassment to another user. If the harassment is confirmed, the examiner performs publication banning or account suspending on the harassing account according to the severity of the harassment.
However, for massive UGC data, the efficiency of manually examining and processing a harassing mention behavior is low. Moreover, this manner of processing after occurrence relies on feedback of a user, and actually many users do not know how to or are unwilling to make a feedback. Therefore, the harassing behavior is often not processed in time, further reducing the efficiency of processing harassment by a mention in UGC.